Steve Howell on how to synthesise both conventional orchestral wind sounds and their rougher, less refined counterparts from the non-Western world.

The principle behind most woodwind sounds is that air is passed down a resonant tube whose apparent length is varied by opening and closing holes along its' length. There are, therefore, two elements to most woodwind sounds - the tone and the breath. This is particularly true of overblown instruments, though on double reed devices such as the oboe and bassoon the breath element is not so prominent.

Oboe is actually one of the most difficult sounds to synthesise accurately, though it is quite easy to recreate the effect of the instrument's reediness. This is done by using as the basic waveform a pulse wave whose pulse width (or mark/space ratio) is asymmetric, ie. in the region of 10:90 (or vice versa), though the actual ratio can be varied according to taste. This should help capture the acoustic sound's inherent thinness. The VCF should be set quite resonant - about halfway - and the cutoff frequency fairly high according to taste. The EG should incorporate a softish attack (not too slow, though) with full sustain and a short release.

Slow (around 4Hz) vibrato can be added if so desired. Experiment with introducing the effect gradually if you wish, though personally I prefer leaving a moderate amount of vibrato on all the time. This should give you a pretty good oboe effect, especially if you smother it in reverb and keep it back in an orchestral mix. Take it down an octave, with suitable tweaking of the VCF cutoff frequency and EG attack, and you should have a reasonable approximation of a bassoon.

If, on the other hand, you feel like being a bit more adventurous, you could try the patch given in Figure 2. Here, a footpedal is used to control the tone of the sustained sound by altering the cutoff frequency and the pulse width slightly as the pedal is moved. If a footpedal is not available, then any DC voltage supply routed via a conveniently located control can be used. You will see also that a sinewave (tuned about two octaves above the pitch of VCO1, with a very rapid attack/decay and no sustain) is used to modulate the VCF very slightly. This element introduces the 'squeak' at the front end of a double reed instrument. You can also, if you wish, sweep the width of the pulse very slightly with the output from a spare EG, as this can create the illusion of reeds 'energising'.

As an option for both of the patches given, you might like to try routing the whole thing through a parametric equaliser with a narrow 'Q' and frequency set around 3K or so. In the absence of a parametric, a graphic equaliser with all but the middle frequency sliders at minimum will perform much the same job, as will the EQ section of your mixer. Finally, if you want to hear an excellent example of a synthesised oboe, have a listen to The War Suite by Gino Vanelli...

Flutes and Clarinets

Clarinets are relatively easy to synthesise, since a straight square wave with a sottish attack should get you close to the acoustic sound without too much trouble. Filter cutoff frequency and ADSR times can be adjusted to suit, and again you could consider the addition of some vibrato. In general, though, all you really need is a good, symmetrical square wave.

The flute, on the other hand, is a different kettle of sonic fish altogether. Whereas oboes and clarinets are shoved in the mouth and blown down, a flautist has to blow over his instrument in much the same way as you or I would blow a bottle. The result is a much more 'breathy' sound, and it's necessary to set up your synth patch in a slightly different way if you wish to recapture that element. Figure 3 gives a patch that can be used to create an effective flute sound: there are two channels - one for the tone and one for the breath. Many people attempt to synthesise the breathiness of a flute simply by mixing a noise generator in with the VCO via the VCF but frankly I've never been a fan of this method: it rarely sounds particularly convincing, due in part to the fact that without splitting the two sound elements into separate parts, complete control over the synthesising process is well-nigh impossible.

The pitched tone portion of the sound is derived from a symmetrical square wave, passed through a VCF whose cutoff frequency is adjusted so that the sound is fairly pure. A triangle wave can be used as an alternative, but steer clear of using a pure sinewave, as it's usually too lacking in harmonics to be heard amongst other instruments. There is a very slight EG sweep of the VCF - just enough to open the envelope a bit at the front end of the sound. The EG controls are set to give a 'blowy' sound, with attack, decay, and release times of about 500ms or thereabouts and the sustain set about halfway. This introduces a subtle touch of harmonic movement, making the overall sound more interesting.

The 'breath' channel is derived from feeding a white noise source (it must be white noise for optimum effect) into a low or bandpass filter which, in turn, is shaped by EG3. The filter can be set more or less as required, with the cutoff frequency and resonance adjusted to suit, while the EG controls can be set to give either a short burst of breath at the front of the note or, with the sustain control up, a hint of 'blow' throughout. Vibrato can be added as desired.

Ethnic Woodwind Instruments

The patches outlined above should give you a fair approximation of orchestral woodwind sounds, and can sound effective either back in the mix or as melody lines.

Other cultures also use similar instruments as a part of their traditional music but, whereas Western civilisations have refined their instruments and accompanying playing techniques, the wind instruments of some ethnic societies have not evolved to such a degree. What this means for the average synthesist is that whilst the wind patches remain essentially the same, variations in some control settings and playing techniques are necessary if these less conventional wind instruments are to be synthesised accurately.

The major difference between the two generations of instruments lies in pitch stability. Arabian, African, and Oriental musicians - without the advantages (?) of a Royal Academy training - place rather less emphasis on holding stable notes for any length of time, with the result that pitch is given more than a passing opportunity to 'wobble'.

To emulate this effect, we can use sample and hold, as patched in Figure 4, instead of a symmetrical sine or triangle wave output from an LFO. In this case, the CV output is routed via a lag time integrator which acts as a slurred, random modulation source. When applied to a VCO in small amounts, this makes the sound more unpredictable and therefore more ethnically realistic and 'human'. Modulation levels are quite critical - make sure you don't go over the top and introduce an effect that's too extreme.

It's worth mentioning at this point that a woodwind player cannot, in fact, actually create true vibrato - the effect produced is amplitude modulation, or tremelo. Having said that, however, when it comes to synthesising wind sounds, pitch modulation is extremely effective, if not absolutely authentic.

Non-Western players often lack the highly-developed embrouchure possessed by orchestral players, and their instruments tend to be more 'breathy' as a result - this is especially true of pan-pipes and African flutes. Adjusting synth patches with this in mind, the most effective modification you can make is to adjust the resonance control so that the filter is almost on the verge of oscillation. You should also try tuning the cutoff frequency so that it bears some relationship to the pitch of the tone. You can adjust the balance between tone and breath to suit the effect you require. Finally, you can experiment with various sets of ADSR times for different effects, though make sure things don't end up sounding utterly unmusical!

Non-Western read instruments can be quite a bit more screechy than our oboes, so some adjustment of the filter controls is necessary for faithful reproduction, along with changes to your EQ settings. You might like to try routing the output of a sample and hold into the pulse width CV input for random pulse width changes at each note: the S/H must be stepped through by the trigger output of the keyboard.

Note-sliding

Many non-Western folk musicians employ a lot of sliding between notes, so portamento can be used in small doses. This should be of the temporary type, provided either by an auto-glide facility on the keyboard (played legato) or by switching it in manually with a pushbutton or footswitch.

Lastly, the music of many non-Western cultures can be quite a bit more dynamic than that of their Western counterparts, so it can be a good idea to route the whole synth sound through a master VCA whose output level is controlled by a footpedal or similar controller. You could even try this technique in the context of Western woodwind sounds as well - and play wind parts with a dynamic range rather wider than that produced by the average Berlin Philharmonic Orchestra flautist!

Putting all these techniques into practice can be something of a headache, requiring as it does a great deal of rehearsal and careful listening to the music of different cultures. It's interesting to note that what a traditional peasant folk musician does almost instinctively requires not only a lot of technology but also a great deal of advance planning and thought on the part of the modern synth player. So much for the wonders of science...

Phrasing

Even though all the points mentioned above should put you on the road to successful wind sound synthesis, your efforts will be all but wasted if you don't pay a reasonable amount of attention to phrasing. Remember, a woodwind player has only a finite amount of breath, so if you start playing very long notes, they're more than likely going to diminish the authenticity of the sound. Try to arrange your synth parts so that there is time for your wind soundalike to 'take a breath', and try to keep them within the range of the instrument you're basing your sound on. Of course, part of the beauty of synthesisers as instruments is that you're rarely restricted by such mechanical considerations, but if realism is your aim, the rules outlined above are simply impossible to ignore.

All wind sounds are essentially monophonic and are therefore best suited to a single-trigger monophonic system - a polyphonic synthesiser might not be capable of creating the sound you want. With the possible exception of the flute, it's pretty rare for woodwind instruments to go divisi, and they're best recorded line-by-line on separate tracks of a multitrack tape. During recording, reverb can be applied as necessary, as can repeat echo if you're looking for a useful special effect. Personally, I find chorus, flanging, and phasing render the sound too 'electronic', though this may of course be precisely the effect you're after - feel free to experiment.

Before we finish, there is one playing technique that can be usefully applied to all the above sounds (or indeed any with a moderately slow attack) and that is pseudo-touch-sensitivity. If a note is played staccato, the attack doesn't have sufficient time to reach its full output level: holding the finger on the key a little longer enables it to do so. So, altering the way you touch the keys can give you quite a wide range of control over the sound's output level, depending on the attack time set.

That, then, concludes our excursion into the world of woodwind sounds. I feel quite strongly that these forms of sound texture are rather underused in much of modern music. Western wind tones can be used as a string section alternative for synth players wanting orchestral size without orchestral cliches, while more exotic sounds can lend a piece of synth-based music a more than welcome 'acoustic' flavour.